Image pre-processing tasks in the spoofing attack prevention journey
Rajesh Kumar (~rajesh37) |
In the facial recognition world, preventing spoofing attack is an uphill task. Most of the existing solution approach like eye blink, 3G image depth approach, illumination , real vs fake eyes support does not yield the expected results. The success factor is less than 30% under the complex face feature scenario. Through this assignment, we take an attempt to prevent spoofing attack through an effective validation using the object detection approach.
What we mean here is, inorder to have a simple and rugged face recognition system, we need to prevent users to spoof the system with a fake image of a person . In this case, as a first step before getting into recognizing the face, the context of the environment is to be studied. If the context has a mobile or attempt to fake the system with an alternate image through mobile, the system should recognize and prevent proceeding with face recognition approach.
The proposal for the Devsprint is as below: 1) Make use of the provided image or create a smaller imageset (Instructions will be given) 2) Images will be of non-standard size , file type and resolution 3) Use python libraries to standardize the images for a given resolution, say 250x250 (Python baseline code will be provided) 4) The pre-processed images will be read by the face identification Model. 5) The error or challenges in feeding the processed images to the model needs to be fixed
We have the liberty to discard the images which does not fit into the requirement as outlier.
I encourage the Beginner and Python experts to attempt solve the challenges of rescaling images to effectively fit into the DL network.
Basic Python programming knowledge Understanding of Image pre-processing Python 3.6 environment
Providing some reference links for the implementation:
Rajesh K Jeyapaul is a Sr.Developer Advocate @ IBM India. In his 23 years of technology journey , he has led various product development projects on JVM technologies, Operating System and cloud services stack. Also, has led various enterprise solution enablement for Banking and Telecom sector. Currently, he is leading enterprise AI solution enablement for Banking clients.
He started his career with Indian Defense Research organization as Research Fellow . He has co-authored various books on performance engineering and has few patents to his credit. He loves mentoring the student and Startup community .
https://www.linkedin.com/in/rajeshjeyapaul/ https://www.youtube.com/watch?v=Tf7cLiZWlgA&t https://www.youtube.com/watch?v=8CDJY0MdW_g&t https://www.youtube.com/watch?v=qlVFMiN5gcU&t=256s